analytics_gen 1.0.5

analytics_gen: ^1.0.5 copied to clipboard

Type-safe analytics from a single YAML plan.

Stop guessing event names. Start treating analytics as code.

analytics_gen generates type-safe Dart code from your analytics YAML files. It ensures your tracking plan, implementation, and documentation are always in sync.

• 🛡️ Type-Safe: No more typos in event names or parameters.
• 📄 Self-Documenting: Your YAML is the single source of truth.
• 🔌 Provider Agnostic: Works with Firebase, Amplitude, Mixpanel, or any custom provider.

Table of Contents #

• Quick Start
• CLI Commands
• Key Features
• Documentation Hub
• Validation & Quality
• Analytics Providers & Capabilities
• Synchronous Logging & Async Providers
• Batch Logging Buffers
• Example
• Testing
• Contributing
• FAQ
• License

Quick Start #

1. Describe events in events/*.yaml (one domain per file). Keep descriptions + parameter docs current.
2. Configure analytics_gen.yaml. The configuration uses a structured format:

   analytics_gen:
     inputs:
       events: 'events'               # Directory path (scans top-level)
       # events: 'events/**/*.yaml'   # Or use glob patterns for recursive scanning
     outputs:
       dart: 'lib/src/analytics/generated'
   

3. Generate code/doc/exports with dart run analytics_gen:generate --docs --exports.
4. Initialize + use the generated API:
   • Singleton: Analytics.initialize(...), then Analytics.instance.logEvent(...).
   • DI: final analytics = Analytics(...), then analytics.logEvent(...).
5. Review diffs for generated Dart, docs, and exports during PRs-treat them as production code.

Need a detailed walkthrough? Head to doc/ONBOARDING.md.

CLI Commands #

dart run analytics_gen:generate              # code only (default)
dart run analytics_gen:generate --docs       # code + docs
dart run analytics_gen:generate --docs --exports
dart run analytics_gen:generate --docs --no-code
dart run analytics_gen:generate --exports --no-code
dart run analytics_gen:generate --validate-only
dart run analytics_gen:generate --plan
dart run analytics_gen:generate --watch              # incremental rebuilds on file change
dart run analytics_gen:audit                         # scan for unused events

Pair these with the configuration you committed to analytics_gen.yaml. Add --no-docs / --no-exports locally if you need a faster iteration loop-the config still drives CI.

Key Features #

• Type-safe analytics - compile-time event + parameter checking, optional allowed-value guards.
• Domain-per-file generation - readable diffs, mixins scoped to a single domain.
• Glob pattern support - use events/**/*.yaml for recursive scanning or organize events in subdirectories.
• Deterministic outputs - sorted parsing with fingerprints to keep reviewers sane.
• Multi-provider fan-out - send the same event to multiple SDKs with error handling.
• Docs + exports - Markdown, CSV, JSON, SQL, SQLite artifacts for stakeholders.
• Runtime plan - Analytics.plan exposes the parsed plan at runtime for debugging or feature toggles.
• Extensible Metadata - attach arbitrary key-value pairs (e.g., owner) to events and parameters in YAML, propagated to code, docs, and exports.
• Enhanced CSV Export - generates multiple CSV files (events, parameters, metadata, relationships) for deep analysis.
• Parameter Validation - define validation rules (regex, length, range) in YAML and get runtime checks automatically.
• Generation Telemetry - track generation performance with built-in metrics (domain processing times, total duration, file counts).
• Capability Discovery - auto-generated documentation shows available capabilities and usage examples in generated code.

Parameter Validation #

Define validation rules directly in your YAML to ensure data quality at the source. The generator creates runtime checks that throw ArgumentError if constraints are violated.

search_event:
  description: User searched for content.
  parameters:
    query:
      type: string
      min_length: 3
      max_length: 100
      regex: "^[a-zA-Z0-9 ]+$"
    result_count:
      type: int
      min: 0
      max: 50

Supported rules:

• regex: Regular expression pattern (String).
• min_length: Minimum string length (int).
• max_length: Maximum string length (int).
• min: Minimum numeric value (num).
• max: Maximum numeric value (num).

Enums for Allowed Values #

If a string parameter defines allowed_values, the generator automatically creates a Dart enum to enforce type safety.

login:
  parameters:
    method:
      type: string
      allowed_values: ['email', 'google', 'apple']

Generated code:

enum AnalyticsAuthLoginMethodEnum {
  email('email'),
  google('google'),
  apple('apple');
  // ...
}

void logAuthLogin({required AnalyticsAuthLoginMethodEnum method}) { ... }

Type-Safe Dart Enums (dart_type) #

You can also use existing Dart enums (or other classes) directly by specifying dart_type. The generator will use this type in the method signature and serialize it using .name.

feature:
  interact:
    parameters:
      status:
        dart_type: VerificationStatus # Must be available in scope
      reason:
        dart_type: ReasonEnum?

Generated code:

void logFeatureInteract({
  required VerificationStatus status,
  ReasonEnum? reason,
}) {
  logger.logEvent(
    name: "feature: interact",
    parameters: {
      "status": status.name,
      if (reason != null) "reason": reason?.name,
    },
  );
}

Note: By default, the generator does not know where VerificationStatus is defined. You must ensure it is imported.

Managing Imports #

You have two ways to add imports for your custom types:

1. Global Imports (Recommended for Barrel Files)

Add a list of imports to your analytics_gen.yaml. These will be added to every generated event file. This is ideal if you have a single "barrel file" that exports all your analytics enums.

analytics_gen:
  inputs:
    imports:
      - 'package:my_app/analytics/analytics_types.dart'
      - 'package:my_app/core/models/user_status.dart'

2. Local Imports (For One-offs)

Specify the import directly on the parameter using the import key. This is useful for unique types used in only one place.

status:
  dart_type: VerificationStatus
  import: 'package:my_app/features/auth/verification_status.dart'

Dual-Write Migration #

When migrating from an old event to a new one, you can use dual_write_to to automatically log to both events during the transition period.

auth:
  login_v2:
    description: New login event
    dual_write_to: ["auth.login"] # Logs to 'auth.login' as well
    parameters:
      method: string

The generator will attempt to call the generated method for the target event if it exists in the same domain, ensuring type safety and logic reuse. If the target is in a different domain or parameters cannot be matched automatically, it falls back to a generic logEvent call.

Extensible Metadata #

Attach metadata to events and parameters using the meta key. Use metadata for ownership and other attributes that should travel with the plan but not the runtime code.

user_login:
  description: User logged in successfully.
  meta:
    owner: "auth-team"
    tier: "critical"
  parameters:
    method:
      type: string
      description: The method used to login (email, google, apple).

This metadata is:

• Available at runtime via Analytics.plan.
• Included in the generated Markdown documentation.
• Exported to JSON and CSV for external analysis.

Shared Event Parameters #

Define common parameters in a central file to ensure consistency and reduce duplication.

1. Create a shared parameters file (e.g., events/shared.yaml):

   parameters:
     flow_id:
       type: string
       description: Unique identifier for the user flow.
     user_id:
       type: int
   

2. Configure it in analytics_gen.yaml:

   analytics_gen:
     inputs:
       shared_parameters:
         - events/shared.yaml
     rules:
       # Optional: Enforce that all parameters must be defined in shared.yaml
       enforce_centrally_defined_parameters: false
       # Optional: Prevent redefining shared parameters in events
       prevent_event_parameter_duplicates: true
   

3. Reference in events:

   checkout_started:
     parameters:
       flow_id: # Inherits type and description from shared.yaml
       item_count:
         type: int
   

Contexts & Global Properties #

Contexts allow you to define global properties (like user attributes, device info, or theme settings) that are managed separately from individual events.

1. Define a context file (e.g., events/user_properties.yaml):

   user_properties:
     user_id:
       type: string
       description: Unique identifier for the user
     user_role:
       type: string
       allowed_values: ['admin', 'viewer']
   

2. Register it in analytics_gen.yaml:

   analytics_gen:
     inputs:
       contexts:
         - events/user_properties.yaml
   

3. Use the generated API:

   // The generator creates a mixin and capability interface
   Analytics.instance.setUserId('123');
   Analytics.instance.setUserRole('admin');
   

4. Implement the capability in your provider: The generator creates an interface (e.g., UserPropertiesCapability). Your analytics provider must implement this interface and register it to handle the property updates.

   class MyAnalyticsService extends CapabilityProviderBase implements IAnalytics {
     MyAnalyticsService() {
       // Register the capability implementation
       registerCapability(userPropertiesKey, _UserPropertiesImpl(this));
     }
     // ...
   }
   
   class _UserPropertiesImpl implements UserPropertiesCapability {
     final MyAnalyticsService _service;
     _UserPropertiesImpl(this._service);
   
     @override
     void setUserProperty(String name, Object? value) {
       // Forward to your SDK (e.g. Firebase setUserProperty)
       _service.firebase.setUserProperty(name: name, value: value);
     }
   }
   

Documentation Hub #

• Onboarding Guide - setup checklist, command reference, troubleshooting.

• Validation & Naming - schema, strict naming rules, and error explanations.

• Naming & Configuration Guide - recipes for "Engineer Friendly" vs "Business Readable" naming, templates, and casing.

• Capabilities - why capability keys exist and how to expose provider-specific APIs.

• Migration Guides - playbooks for moving Firebase manual strings, Amplitude events, and Mixpanel plans into YAML while keeping downstream dashboards stable.

• Code Review checklist - what to inspect in YAML, generated code, docs, and provider adapters during PRs.

• Scalability & Performance - benchmarks and limits for large enterprise plans (100+ domains / 1000+ events).

• Performance Guide - detailed optimization tips for build and runtime.

• Troubleshooting - common errors and fixes.

• API Reference - generated Dart API docs on pub.dev. CLI Commands

dart run analytics_gen:generate              # code only (default)
dart run analytics_gen:generate --docs       # code + docs
dart run analytics_gen:generate --docs --exports
dart run analytics_gen:generate --docs --no-code
dart run analytics_gen:generate --exports --no-code
dart run analytics_gen:generate --validate-only
dart run analytics_gen:generate --plan
dart run analytics_gen:generate --watch              # incremental rebuilds on file change
dart run analytics_gen:audit                         # scan for unused events

Pair these with the configuration you committed to analytics_gen.yaml. Add --no-docs / --no-exports locally if you need a faster iteration loop-the config still drives CI.

Validation & Quality #

• Run dart run analytics_gen:generate --validate-only in CI to block invalid YAML before files are written.

• CI Guardrails: Add a step in your CI pipeline to run generation and check for uncommitted changes (git diff --exit-code). This ensures that the generated code, docs, and exports are always in sync with the YAML definitions.

• Naming strategy (analytics_gen.naming) enforces consistent identifiers-override per-field when legacy plans demand it.

• Configurable Event Casing: You can control the format of the generated event string (the value sent to analytics providers) using the casing option in analytics_gen.yaml.

  • snake_case (default): order: completed -> order_completed (Engineer Friendly, great for SQL).
  • title_case: order: completed -> Order Completed (Business Readable, great for Amplitude/Mixpanel).
  • original: Preserves the raw template output (e.g. order: completed).

  analytics_gen:
    naming:
      casing: snake_case # or title_case, original
  

• Strict Event Naming: The parser forbids string interpolation in event names (e.g. View ${page}) to prevent high-cardinality events from polluting your analytics data. This is now enforced by default.

• Docs/JSON/SQL outputs embed a fingerprint derived from the plan; unexpected diffs mean someone skipped regeneration.

• Full details live in doc/VALIDATION.md.

"Dead Event" Audit #

Over time, analytics plans grow, and events are deprecated or removed from the codebase but left in the YAML. The audit command scans your Dart code (lib/) for usages of the generated event methods.

dart run analytics_gen:audit

Output Example:

[WARN] Found 2 dead events (0 usages):
[WARN]   - auth.login_v2 (Method: logAuthLoginV2)
[WARN]     Defined in: events/auth.yaml:17
[WARN]   - purchase.cancelled (Method: logPurchaseCancelled)
[WARN]     Defined in: events/purchase.yaml:19

Supported flags:

• --config: Path to your config file (default: analytics_gen.yaml).
• --verbose: Show detailed logs.

Analytics Providers & Capabilities #

The generated mixins call logger.logEvent(...). You are in control of the underlying providers:

• Singleton Usage: Analytics.initialize(provider) and access via Analytics.instance.

• Dependency Injection: Create an instance final analytics = Analytics(provider) and inject it where needed.

• Implement IAnalytics for a single SDK.

• Use MultiProviderAnalytics to fan out events while isolating provider failures.

• Need provider-specific hooks (user properties, timed events, etc.)? Register typed capabilities via CapabilityKey<T> (extend CapabilityProviderBase to wire a registry automatically). Consumers request them when needed, keeping the base interface small. See doc/CAPABILITIES.md for a junior-friendly walkthrough.

Mock + async adapters (MockAnalyticsService, AsyncAnalyticsAdapter) are included for tests and await-heavy flows.

Synchronous Logging & Async Providers #

IAnalytics.logEvent stays synchronous on purpose:

• UI instrumentation should be fire-and-forget so frames are never blocked by analytics I/O.
• Generated methods are widely sprinkled across apps; forcing await would leak runtime coupling into feature code.
• Tests remain simple because MockAnalyticsService records events immediately.

When a provider must await network flushes or heavy batching, queue that work behind an adapter instead of changing the base interface. A reference pattern:

final class QueueingAnalytics implements IAnalytics {
  QueueingAnalytics(this._asyncAnalytics);

  final IAsyncAnalytics _asyncAnalytics;
  final _pending = <Future<void>>[];

  @override
  void logEvent({required String name, AnalyticsParams? parameters}) {
    _pending.add(
      _asyncAnalytics.logEventAsync(name: name, parameters: parameters),
    );
  }

  Future<void> flush() => Future.wait(_pending);
}

final asyncProvider = AsyncAnalyticsAdapter(
  MultiProviderAnalytics([
    FirebaseAnalyticsService(firebase),
    AmplitudeService(amplitude),
  ]),
);

Analytics.initialize(QueueingAnalytics(asyncProvider));

• The UI still calls synchronous generated methods.
• QueueingAnalytics tracks outstanding futures so diagnostics/tests can await flush().
• AsyncAnalyticsAdapter wires into existing synchronous providers, which keeps compatibility with code gen.

This same pattern appears in example/lib/main.dart, where the async adapter is exercised alongside the synchronous runtime.

Batch Logging Buffers #

When you need to control network fan-out-slow uplinks, cellular metering, or provider SDKs that enforce batch delivery-wrap your async provider with BatchingAnalytics.

final asyncAdapter = AsyncAnalyticsAdapter(
  MultiProviderAnalytics([
    FirebaseAnalyticsService(firebase),
    AmplitudeService(amplitude),
  ]),
);

final batching = BatchingAnalytics(
  delegate: asyncAdapter,
  maxBatchSize: 25,
  maxRetries: 3, // Drop events after 3 failures (poison pill protection)
  flushInterval: const Duration(seconds: 5),
  onFlushError: (error, stack) {
    print('Batch flush failed: $error');
  },
  onEventDropped: (name, params, error, stack) {
    print('Event dropped after max retries: $name');
    // Optional: Save to disk (Dead Letter Queue)
  },
);

Analytics.initialize(batching);

// ... app runs, events buffer automatically.

Future<void> onAppBackground() async {
  await batching.flush(); // or await batching.dispose();
}

• maxBatchSize forces a flush when enough events accumulate.
• flushInterval (optional) drains on a timer for long-lived sessions.
• flush() returns a Future so lifecycle hooks (AppLifecycleState.paused, integration-test teardown) can wait for delivery.
• Exception Handling: Manual calls to flush() will throw if retries are exhausted, allowing you to handle network failures. Automatic background flushes catch errors and report them to onFlushError without crashing the app.

Combine BatchingAnalytics with AsyncAnalyticsAdapter to await multiple providers without changing the generated, synchronous API surface.

Example #

The example/ directory shows a working tracking plan + app integration.

cd example
dart pub get
dart run analytics_gen:generate --docs --exports
dart run lib/main.dart

Generated artifacts inside the example mirror what your app will emit. Use it as a sandbox before editing your production plan. Prefer running the Flutter sample via flutter run so you can tap through the buttons and inspect recorded events live-the UI simply calls into a controller that wraps the generated mixins so logging stays out of widgets.

Testing #

• Unit tests should initialize Analytics with MockAnalyticsService (or the async adapter) and assert on recorded events.
• Use Analytics.reset() in tearDown to clear the singleton instance between tests.
• Add dart run analytics_gen:generate --validate-only to CI so schema errors fail fast.
• Run dart analyze + dart test before committing-analytics code follows the same standards as the rest of your Flutter/Dart app.

Generated Test Matchers #

You can opt-in to generate typed Matcher helpers for package:test by enabling test_matchers in your configuration:

analytics_gen:
  targets:
    test_matchers: true # Generates test/analytics_matchers.dart

This generates is{EventName} matchers that simplify verifying logEvent calls:

import 'analytics_matchers.dart';

test('logs login event', () {
  // ... trigger action ...
  
  verify(() => analytics.logEvent(
    name: 'auth.login', 
    parameters: isAuthLogin(
      method: AuthLoginMethod.email,
    ),
  ));
});

Contributing #

Contributions welcome! Please:

1. Fork the repo and open a PR with focused changes.
2. Add unit tests for new features/validations.
3. Run dart analyze, dart test, and the generator before submitting.
4. Document user-facing changes in the README or the relevant doc under doc/.

Every PR description flows through .github/pull_request_template.md, which links directly to doc/CODE_REVIEW.md. Walk through that checklist (YAML, generated Dart, doc/exports, providers, and security) before requesting a review so reviewers only need to validate, not rediscover, regressions.

FAQ #

• Why YAML instead of defining events directly in Dart? YAML is tooling-agnostic and easy for product/analytics teams to edit. analytics_gen transforms it into code, docs, and exports without duplication.

• Can I migrate existing events? Yes. Start by mirroring your current events in YAML, run the generator, and gradually replace hard-coded logEvent calls with generated methods.

• Does this lock me into one provider? No. Implement or combine multiple providers. Capabilities expose provider-specific features without bloating the global interface.

• Can I change the event naming style (e.g. to "Title Case")? Yes. Use analytics_gen.naming.casing in analytics_gen.yaml. See the Naming Guide for examples.

• Is it safe to commit generated files? Committing generated files is supported. Outputs are deterministic and intended to be reviewed in PRs (see doc/CODE_REVIEW.md).

License #

Licensed under the Apache License, Version 2.0. See LICENSE for details.